14b892482bc4d3744f01412593588515.jpg

Contact


🌍 Champaign, IL

<aside> 📩 **s[email protected]**

</aside>

<aside> <img src="attachment:16c7ec70-e7da-4d4f-bfb6-db1d2ac77393:128px-Google_Scholar_logo.svg.png" alt="attachment:16c7ec70-e7da-4d4f-bfb6-db1d2ac77393:128px-Google_Scholar_logo.svg.png" width="40px" /> Google Scholar

</aside>

<aside> <img src="attachment:0852e7a0-6246-40d6-9cca-4c13516da734:icons8-twitter-500.svg" alt="attachment:0852e7a0-6246-40d6-9cca-4c13516da734:icons8-twitter-500.svg" width="40px" /> X

</aside>


Misc.

I like minions, watching stand-up comedy, cars and planes.


Acknowledgement

I borrowed this template from my amazing lab-mate Deema.

🍟 About Me


I am Dylan, a Ph.D. Student in Computer Science at the University of Illinois Urbana-Champaign (UIUC)🌽 , advised by Prof. Hao Peng. A proud member of our lab (as of the most recent update, our lab’s official name is ALTA).

I’m broadly interested in language model post-training. My previous work has focused on the data-centric aspects of post-training, and moving forward, I’m eager to explore new directions such as offline-to-online reinforcement learning, incentivizing proactive reasoning behaviors for knowledge discovery, and enabling the rapid deployment of foundation model–based agents in novel environments.

I’m also deeply excited about the growing role of foundation models in scientific discovery—and the rich post-training algorithmic challenges that come with it.

I am looking for Research Internship opportunities for 2026 summer

🎓 Education


uiuc-logo.png

Ph.D. in Computer Science

University of Illinois Urbana-Champaign 🌽 🐿️

Advisor: Prof. Hao Peng

2022 - Present

🔔 News


📰 Selected Works


2025

The Best Instruction-Tuning Data are Those That Fit

Improving Influence-based Instruction Tuning Data Selection for Balanced Learning of Diverse Capabilities

Entropy-Regularized Process Reward Model

ScaleBiO: Scalable Bilevel Optimization for LLM Data Reweighting


2024

Only-IF: Revealing the Decisive Effect of Instruction Diversity on Generalization

SciCode: A Research Coding Benchmark Curated by Scientists


2023

Making Large Language Models Better Reasoners with Step-Aware Verifier


2021

Pre-training Co-evolutionary Protein Representation via a Pairwise Masked Language Model

💼 Internship Experience


Google_2015_logo.svg.webp

Microsoft-Logo.png

Microsoft-Logo.png

Student Researcher

Google

Research Intern

Microsoft Research

Research Intern

Microsoft Research


Last updated: Sept, 2025